NewChat Models

NVIDIA

Nemotron 3 Super

Balanced Nemotron workhorse for text, coding and tool-assisted workflows, using a paid 256K OpenRouter route.

BalancedTool useText
Open AppBrowse All Models

At a glance

Know the model before you prompt.

OpenRouter API specifications
Context window
262,144 tokens
Maximum output
Provider-dependent
Inputs → output
Text → Text
Knowledge cutoff
Not verified

API model ID: nvidia/nemotron-3-super-120b-a12b

Capabilities & boundaries

What it supports. Where the limits are.

Tool support requires the appropriate API integration; a supported tool is not automatically active in every chat.

Supported API features and tools

  • A 120B-total, 12B-active hybrid mixture-of-experts model in NVIDIA’s official specification
  • Text-focused conversational and coding workflows, including synthesis over supplied reference material
  • Reasoning support with medium and low effort listed in OpenRouter’s model metadata
  • Tool calling and structured-output workflows through compatible hosting integrations

Before you choose

  • Collaborative-agent capability is not a built-in team of agents in every app. Orchestration, external tools and execution permissions belong to the integration.
  • NVIDIA reports pre-training data through June 2025 and post-training through February 2026. These are different data stages, not one guaranteed knowledge cutoff or live search capability.
  • The OpenRouter catalog reference is 262,144 tokens (256K). NVIDIA’s underlying model card describes up to 1M in suitable deployments; that is not the app’s confirmed route limit. Individual hosts can impose smaller windows. Leave room for the reply and reasoning.
  • This route accepts text and returns text, not image, audio or video input/output. Provide extracted text for document work; a prompt cannot add vision support.
  • Tool calls and structured outputs require a compatible host and integration. The model does not independently browse the web, execute code or access private files.
  • Answers, extracted fields and generated code can be wrong. Supply evidence and acceptance criteria, check cited passages and test code before use. A long context does not guarantee complete recall.

Choose the reasoning effort

lowmedium · default

OpenRouter’s catalog lists reasoning enabled by default with medium effort and a low option. Test low on bounded tasks before increasing the budget. Confirm the host’s accepted parameters and the app’s exposed controls; do not assume Ultra’s high level or a self-hosted chat-template flag applies to Super’s routed interface.

  • Use nvidia/nemotron-3-super-120b-a12b for the owner-confirmed paid route. NVIDIA is the model developer; OpenRouter routes the request to a hosting provider.
  • DekaLLM lists 262,144 context and 235,929 maximum output tokens, while DeepInfra lists the same context with 16,384 maximum output. Prompt length and reasoning reduce the usable answer budget.
  • The reviewed DekaLLM endpoint lists structured_outputs; DeepInfra’s supported-parameter list is different. A JSON-looking response is not a substitute for schema validation.

Put it to work

Start with a more useful prompt.

Original examples from EZ Ai Assist. Adapt these to your task and the features available in your workspace.

Workflow 01

Triage a support queue

Classify recurring work with explicit escalation rules.

Classify the tickets below using the provided categories, severity definitions and support policies. Return ticket ID, category, severity, supporting phrase and next action. Mark ambiguous cases for review. Draft a brief response only when the policy supports it; do not promise refunds, change account settings or invent missing customer facts.

Workflow 02

Review a bounded feature patch

Focus review on agreed behavior and testable regressions.

Review this patch against the acceptance criteria. Check input validation, state transitions and error handling. For each concrete finding, cite the affected function, explain the failure scenario and propose one targeted test. Keep unrelated style changes out of scope and clearly label tests as proposed unless their results are supplied.

Workflow 03

Extract a project action register

Convert a document bundle into accountable next steps.

Read these meeting notes and project documents. Produce an action register containing task, owner if explicitly named, due date if stated, source section and dependency. Separate approved actions from suggestions and mark missing fields as not stated. End with contradictions that need a project owner’s decision.

Developer reference

OpenRouter API pricing

These are named-provider OpenRouter API reference prices, not EZ Ai Assist subscription prices.

View EZ Ai Assist plans
DekaLLM route reference on OpenRouter · USD per 1,000,000 tokens
Token typePrice
Input$0.08
Output$0.45
  • Rates were checked October 8, 2026. The named host is a pricing reference, not confirmation that the app pins that host. Routing, availability, cache eligibility and promotions can change the actual bill.
  • DekaLLM lists $0.08 input and $0.45 output per million tokens. This table does not assume a cache rate that the reviewed endpoint did not publish.
  • DeepInfra instead lists $0.085 input and $0.40 output. A lower input rate is not necessarily a lower total bill for an output-heavy task.
  • For a rough uncached estimate, multiply input tokens by the input rate and output tokens by the output rate, then divide by one million. Inspect the provider’s usage and billing records for reasoning and any additional charges.

Common questions

A few things worth knowing.

What makes Super a workhorse option?

It is positioned for conversational, agentic and high-volume text workloads, including coding and retrieval-assisted answers. Evaluate whether it meets your own quality threshold before moving to Ultra or down to a compact model.

How is Super different from Ultra?

The official cards describe Super as 120B total and 12B active parameters, versus Ultra’s 550B total and 55B active. Their OpenRouter effort levels and host rates differ too. Those specifications describe the models, not guaranteed results on your tasks.

Does Super automatically run tools or other agents?

No. It can generate tool requests, but an application must supply tools, execute permitted actions and return results. Use explicit approval boundaries for actions with external effects, and validate tool arguments before execution.

Which output ceiling should I use?

Use the selected endpoint’s limit, not the largest number shown anywhere in the catalog. DekaLLM lists 235,929 output tokens and DeepInfra 16,384, both with 262,144 context. Account for the prompt and reasoning within the total budget.

Are these the app’s confirmed model IDs?

Yes. The app owner confirmed these exact paid OpenRouter IDs, with text-only 256K routes and no :free variants. That confirmation does not test a live request or guarantee that every hosting feature and reasoning control is exposed in your workspace.

Are the token prices EZ Ai Assist subscription prices?

No. These are named-host developer reference rates on OpenRouter. Hosting choices, discounts and supported features can change; your EZ Ai Assist plan and any limits displayed in the app are separate.

Can I use this model to inspect an image or search the web?

Not from the model alone. These routes are text-only. Supply extracted text or use a verified vision model for images. Web search and other tools require an enabled integration; a generated tool request is not proof an action happened.

Check the source

Official documentation

Specifications and API prices checked on . Example prompts and workflow advice are editorial guidance from EZ Ai Assist.

Same provider

More from NVIDIA

NVIDIA

Nemotron 3.5 Super VL

NVIDIA’s vision-language reasoner for text, images and visual documents. The app uses paid OpenRouter hosting; context limits vary by serving host.

NVIDIA

Nemotron 3 Ultra

NVIDIA’s flagship Nemotron reasoner for complex text, code and multi-step analysis, using a paid 256K OpenRouter route.

NVIDIA

Nemotron 3.5 Lightning

Efficiency-focused Nemotron model for low-cost text workflows and rapid iteration, using a paid 256K OpenRouter route.

NVIDIA

Nemotron 3 Nano

Compact Nemotron model for short replies, extraction and bounded reasoning, using a paid 256K OpenRouter route.